DepthMode and reconstruct position using linear depth

I’m trying to reconstruct position using linear depth by using this technique. From the article, he render linear depth by using z (in view space) divide by FarClip.

float4 DepthPS(in float in_fDepthVS : TEXCOORD0) : COLOR0 { // Negate and divide by distance to far-clip plane // (so that depth is in range [0,1]) // This is for right-handed coordinate system, // for left-handed negating is not necessary. float fDepth = -in_fDepthVS/g_fFarClip; return float4(fDepth, 1.0f, 1.0f, 1.0f); }
Urho already had function to render linear depth but using dot product of clipPos.zw and DepthMode.zw. As seen on this Transform.glsl

float GetDepth(vec4 clipPos) { return dot(clipPos.zw, cDepthMode.zw); }
My question is: What actually the use of DepthMode? Is that any difference from that article implementation?

DepthMode is basically just a dot product selector for whether the real Z-value is in the Z or W coordinate. It changes based on whether the camera is orthographic or perspective. You get a value between 0 (nearclip plane) and 1 (farclip plane) from calling GetDepth() in the VS, so you can just pass the value directly to PS.

Thanks for the answer.
I’m still confused about depth mode but after hours of trying, I found out I can simply reconstruct view space position using linear depth by multiplying linear depth to farPlane. The article’s technique is works on Urho linear depth data.
Here’s the code:

[code]// Screen Position
vec2 screenPos = vScreenPos.xy / vScreenPos.w; //vScreenPos is GetScreenPos(gl_Position) on vertexshader
vec2 nScreenPos = screenPos * 2.0 - 1.0;

// Get far plane position at View Space
vec3 farPlane = vec3(vFrustumSize.xy * nScreenPos, vFrustumSize.z); // vFrustumSize is cFrustumSize on vertexshader

// Get linear depth
vec3 linearDepth = DecodeDepth(texture2D(sDepthBuffer, screenPos).rgb);

// Reconstruct view space position by using linear depth
vec3 viewPosition = linearDepth * farPlane;
[/code]